Reciprocal interactions between mitral and granule cells in the olfactory bulb external plexiform layer (EPL) modify the timing of mitral cell action potentials and thereby influence the information that the olfactory bulb exports to its multiple targets. Theoretical and experimental results produced by each of the two collaborating PIs of this proposal have shown that this powerful and plastic EPL network is not responsible for the simple "contrast-enhancement" function often attributed to it (instead, this largely occurs in the glomerular layer). Rather, EPL interactions perform additional, subsequent operations on odor representations that mediate changes in odor perception and representation based in part on an individual animal's history of odor learning. Specifically, we here propose that computations in the EPL serve to render mitral cell output patterns selective for certain higher-order features of odors in the same sense that neurons in primary visual cortex are selective for higher-order visual features such as edge and orientation - that is, they reflect the co-activation of certain spatiotemporal combinations of receptors that together are characteristic of a meaningful odor. We here propose to develop detailed theoretical models of this hypothesis and its implications and to test its main critical predictions experimentally. The model in its present form predicts (1) that whereas granule cells can be excited by mitral cell lateral dendrites irrespective of their physical proximity, spike timing in mitral cells is affected only by inhibition from physically neighboring granule cells, and (2) that granule cells require the simultaneous activation of specific sets of afferent (glomerular) inputs in order to evoke a whole-cell regenerative response and thereby evoke lateral inhibition. This architecture has substantial implications for the processing of odor representations that we will develop in Aim 1. To test this model, we will determine whether granule cell effects on mitral cell activity depend on physical proximity using spatially selective optogenetic activation and silencing of granule cells (Aim 2), measure the form and specificity of the afferent activity patterns required to evoke spikes in granule cells using optical stimulation of olfactory bulb glomeruli (Aim 3), and test the model's assumptions regarding the structure of olfactory bulb plasticity by measuring the perceptual effects of competing odor representations (Aim 4). The intellectual merit of this application derives from its use of state-of-the-art computational modeling to structure the proposed experiments and interpret their results, along with the use of newly-developed experimental techniques to address the longstanding questions about EPL function and processing that the theory described herein has framed and rendered testable. The collaboration between PIs Cleland and Schaefer is essential to the success of this proposal, as PI Schaefer's experimental techniques are uniquely able to test the prediction of PI Cleland's theoretical models. The efficiency of this collaboration is enhanced by the cross-competence of the PIs: PI Schaefer is competent in both computational and behavioral approaches and utilizes both in his research, whereas PI Cleland is competent in electrophysiological approaches and utilizes them in his research. This collaboration will benefit students and postdocs at both institutions by integrating them into a genuinely interdisciplinary framework encompassing both experimental and computational approaches, and facilitating their cross-training by enabling travel between labs. Consequently, the broader impacts of this proposal include the cross-training of students from diverse backgrounds in coordinated theoretical and experimental techniques as well as exposing them to both American and German laboratories. Both PIs have a strong history of training undergraduates and women in areas in which women remain underrepresented. This proposal also provides for the substantial participation of undergraduate researchers.